Multiframe fusion of undersampled 3D imagery

Abstract

3D imaging LADAR Avalanche Photo-Diode (APD) arrays are emerging as an important technology for future generation remote sensing and reconnaissance systems. One important limiting factor in their development is the pixel pitch, which is typically 100 micrometers. This makes the spatial resolution of an APD array at least 10 times lower than their Charge Coupled Device (CCD) array counterparts. With the development of smaller pixel-pitch APD arrays in 3D applications that rival those of CCD's used in 2D imaging many years away, this research endeavors to fuse 3D images with poor spatial resolution with 2D images to obtain superior spatial and range resolution.Microscanning or dithering is a technique used to combine multiple images of the same scene to obtain a more densely sampled version. This techniques has been applied successfully to both 2D and 3D imagery. In this paper, a standard microscanning approach is compared to a new technique involving the use of a properly sampled 2D image combined with multiple low-resolution 3D images in order to obtain a more densely sample 3D image. The comparison will report the mean squared range error obtained from the traditional microscanning approach and the new 2D/3D fusion approach as a function of the number of 3D frames used in the image reconstruction. Since the 2D image contains no range information, improvements in mean squared range error by the new algorithm demonstrate its ability to successfully fuse 2D add 3D information.